[go-nuts] [ANN] Go-gt (Graph Theory) updated

It would be nice if Dijkstra and the others accepted an interface type with
required access methods (and could internally special-case Matrix). Many of
the graph problems I'd need to solve are very sparse, and an n^2 allocation
may be a large waste when there may only be (n,nlogn] connections,
especially if n is large.

I have a graph package which is incomplete but works for what I need
that you may be interested in. It's reasonably fast and makes use of a
Node/Edge interface model using an edge list representation. It the
moment it does not handle extensive node deletion/replacement well
unless you perform compaction (which will break node ID consistency).

code.google.com/p/biogo.graph

Dan

would anyone like to participate? my skills are limited and style is poor,
as you can see :-(

++pac

--